Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals

Page created by Claude Kim
 
CONTINUE READING
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                 1

        Flight Demonstration of High Altitude Aircraft
               Navigation with Cellular Signals
 Zaher M. Kassas, Joe Khalife, Ali Abdallah, Chiawei Lee, Juan Jurado, Steven Wachtel, Jacob Duede, Zachary
                       Hoeffner, Thomas Hulsey, Rachel Quirarte, and RunXuan Tay

   Abstract—This article presents the first demonstration of                       leading to the loss of the main and auxiliary GPS units in
navigation with cellular signals of opportunity (SOPs) on a high                   some cases. What is alarming is the increasing trend of GPS
altitude aircraft. An extensive flight campaign was conducted                      interference– the majority of the aforementioned incidents
by the Autonomous Systems Perception, Intelligence, and Nav-
igation (ASPIN) Laboratory in collaboration with the US Air                        took place since 2019. What is more, previously undisclosed
Force (USAF) to sample ambient downlink cellular SOPs in                           Federal Aviation Administration (FAA) data for a few months
different regions in Southern California, USA. Carrier phase                       in 2017 and 2018 detail hundreds of aircraft losing GPS
measurements were produced from these signals, which were                          reception. On a single day in March 2018, 21 aircraft reported
subsequently fused in an extended Kalman filter (EKF) along                        GPS problems to air traffic controllers near Los Angeles,
with altimeter measurements to estimate the aircraft’s state
(position, velocity, and time). Three flights are performed in three               California, USA [2]. These and other incidents uncover the
different regions: (i) rural, (ii) semi-urban, and (iii) urban. A                  vulnerabilities of existing aircraft navigation systems, which
multitude of flight trajectories and altitudes above ground level                  are highly dependent on global navigation satellite system
(AGL) were exercised in the three flights: (i) a 51 km trajectory                  (GNSS) signals and their augmentation systems (e.g., ground-
of grid maneuvers with banking and straight segments at about                      based augmentation system (GBAS) and space-based aug-
5,000 ft AGL, (ii) a 57 km trajectory of a teardrop descent from
7,000 ft AGL down to touchdown at the runway, and (iii) a                          mentation system (SBAS)) [3], [4]. There is an urgent need
55 km trajectory of a holding pattern at about 15,000 ft AGL.                      for complementary robust and accurate navigation systems to
The estimated aircraft trajectory is computed for each flight                      ensure aviation safety.
and compared with the trajectory from the aircraft’s on-board                         In 2019, the International Civil Aviation Organization
navigation system, which utilized a GPS receiver coupled with                      (ICAO) issued a Working Paper titled “An Urgent Need to
an inertial navigation system (INS) and an altimeter. Cellular
SOPs produced remarkable sustained navigation accuracy over                        Address Harmful Interferences to GNSS,” where it concluded
the entire flight trajectories in all three flights, achieving a three-            that harmful RFI to GNSS would prevent the full continuation
dimensional position root mean-squared error (RMSE) of 10.53                       of safety and efficiency benefits of GNSS-based services.
m, 4.96 m, and 15.44 m, respectively.                                              Moreover, there was a call for supporting multi-disciplinary
                                                                                   development of alternative positioning, navigation, and timing
                                                                                   strategy and solutions to complement the use of GNSS in
                           I. I NTRODUCTION
                                                                                   aviation [5]. In 2021, the U.S. Department of Transportation
   A quick search of the phrase “Global Positioning System                         released the “Complementary Positioning, Navigation, and
(GPS)” on the aviation safety reporting system (ASRS) re-                          Timing (PNT) and GPS Backup Technologies Demonstration
turns 579 navigation-related incidents since January 2000. The                     Report” to Congress. The report concluded that while there
ASRS is a publicly available reporting system established by                       are suitable, mature, and commercially available technologies
NASA to identify and address issues reported by frontline                          to back up or to complement GPS, none of these systems
personnel in the aviation system [1]. A deeper look at the                         alone can universally back up the PNT capabilities provided
data reveals that out of these 579 incidents, a malfunction or                     by GPS and its augmentations, necessitating a diverse universe
failure was detected in navigation sensors with the following                      of PNT technologies [6]. Moreover, in 2021, the National
occurrences: 508 in “GPS & Other Satellite Navigation,” 34                         Institute of Standards and Technology (NIST) issued a report
in “Navigational Equipment and Processing,” 14 in “Flight                          on “Foundational PNT Profile: Applying the Cybersecurity
Dynamics Navigation and Safety,” 12 in “Altimeter,” and 6 in                       Framework for the Responsible Use of PNT Services,” where
“Positional/Directional Sensing.” Among these incidents, 100                       it identified signals of opportunity and terrestrial RF sources
are suspected to be due to GPS jamming and interference,                           (e.g., cellular) as a mitigation category that apply to the PNT
                                                                                   profile [7].
   This work was supported in part by the Office of Naval Research (ONR)
under Grant N00014-19-1-2511 and Grant N00014-19-1-2613; in part by                   Among terrestrial RF signals of opportunity (SOPs), cellular
Sandia National Laboratories under Award 1655264, in part by the National          signals have shown tremendous potential as an alternative PNT
Science Foundation (NSF) under Grant 1929965; and in part by the U.S.              source [8] due to their inherently desirable attributes:
Department of Transportation (USDOT) under Grant 69A3552047138 for the
CARMEN University Transportation Center (UTC).                                        • Abundance: cellular base stations are abundant in most
   Z. Kassas, J. Khalife, and A. Abdallah are with the University of California,         locales, with the number of base stations slated to in-
Irvine. C. Lee, J. Jurado, S. Wachtel, J. Duede, Z. Hoeffner, T. Hulsey, and             crease dramatically with future cellular generations.
R. Quirarte are with the US Air Force, USA. R. Tay is with the Republic
of Singapore Air Force, Singapore. (Corresponding author: Z. Kassas, email:           • Geometric diversity: cellular base stations are placed in
zkassas@ieee.org).                                                                       favorable geometric configurations by construction of the
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                         2

      cellular infrastructure.                                      of samples of 3G code division multiple access (CDMA) and
  •   Frequency diversity: in contrast to GNSS signals, cellular    4G LTE signals were recorded under various conditions.
      signals are transmitted at a wide range of frequencies,          This article provides the first extensive demonstrations of
      which makes more difficult to be simultaneously jammed        their kind of utilizing cellular SOPs for navigation purposes
      or spoofed.                                                   on high altitude aircraft. The aim of these demonstrations
   • High received power: received cellular carrier-to-noise        is to show that should GNSS signals become unavailable or
      (CNR) ratio is commonly tens of dBs higher than that          unreliable mid-flight, cellular SOPs could be used to produce
      of GNSS signals, even in deep urban canyons and indoor        a sustained and accurate navigation solution over trajectories
      environments [9].                                             spanning tens of kilometers.
   • High bandwidth: downlink cellular signals can be up to            To demonstrate the feasibility of aircraft navigation with
      20 MHz (in 4G long-term evolution (LTE)) and even             cellular SOPs, three flights are performed in three different
      higher in future generations, which yields precise time-      regions: (i) rural, (ii) semi-urban, and (iii) urban. A multitude
      of-arrival (TOA) estimates.                                   of flight trajectories and altitudes above ground level (AGL)
   • Free to use: the cellular infrastructure is already oper-      were exercised in the three flights: (i) a 51 km trajectory of
      ational; thus, with specialized receivers, navigation ob-     grid maneuvers with banking and straight segments at about
      servables (pseudorange, carrier phase, and Doppler) can       5,000 ft AGL, (ii) a 57 km trajectory of a teardrop descent
      be extracted from the “always on” transmitted signals.        from 7,000 ft AGL down to touchdown at the runway, and
   Recent results have shown the ability of cellular SOPs           (iii) a 55 km trajectory of a holding pattern at about 15,000 ft
to yield meter-level-accurate navigation on ground vehicles         AGL.
[10], [11] in urban environments and submeter-level-accurate           The aircraft’s trajectory is estimated for each flight exclu-
navigation on UAVs [12], [13]. Moreover, the robustness and         sively from cellular SOPs using an extended Kalman filter
availability of cellular SOPs have been demonstrated in a GPS-      (EKF). The estimated aircraft trajectory is compared with the
jammed environment [14].                                            aircraft’s on-board navigation system, which used a GPS-aided
   Assessing cellular signals for aerial vehicles have been the     inertial navigation system (INS) and an altimeter. Cellular
subject of several studies recently [15]. These studies span        SOPs produced remarkable navigation accuracy in all three
radio channel modeling [16], [17]; evaluation of signal quality     flights, achieving a position root mean-squared error (RMSE)
in terms of received signal power [18], [19], interference from     of 10.53 m, 4.96 m, and 15.44 m, respectively.
cellular transmitters [20]–[22], and coverage and connectivity         The rest of this article is organized as follows. Section II
[23]–[25]; and standards recommendations [26], [27]. How-           describes the aircraft dynamics and cellular SOP measurement
ever, the majority of these studies focused on evaluating cel-      model. Section III formulates the EKF navigation framework.
lular signals for communication purposes with little attention      Section IV describes the experimental setup with which the
to evaluating them for navigation purposes [28]. Moreover,          aircraft was equipped and overviews the environments in
they considered UAVs flying at low altitudes (up to 500 ft)         which the flight campaigns took place. Section V presents ex-
and slow speeds (up to 50 km/h). A recent study revealed that       perimental aircraft navigation results exclusively with cellular
cellular signals can be acquired and tracked at altitudes as        signals. Section VI gives concluding remarks.
high as 23,000 ft above ground level (AGL) and at horizontal
distances of more than 100 km from cellular transmitters                               II. M ODEL D ESCRIPTION
[29]. However, the potential of cellular SOPs for high altitude
aircraft navigation has not been thoroughly assessed. This            This section describes the aircraft dynamics and cellular
article aims to perform the first assessment of cellular SOPs for   SOP measurement models used in the rest of the article.
aircraft navigation by addressing the following question: Can
cellular SOPs be received and exploited at aircraft altitudes to    A. Aircraft Dynamics Model
produce a robust navigation solution?                                  Depending on the aircraft’s motion and sensor suite, differ-
   To answer this question, an unprecedented aerial flight          ent dynamic models can be used to describe its dynamics. The
campaign was conducted in March 2020 by the Autonomous              goal of this article is to assess the minimum performance of
Systems Perception, Intelligence, and Navigation (ASPIN)            aircraft navigation with cellular SOPs exclusively. As such, a
Laboratory in collaboration with the United States Air Force        simple, yet effective continuous Wiener process acceleration
(USAF) at the Edwards Air Force Base (AFB), California,             model is employed, which upon discretization at a constant
USA. The cellular software-defined radios (SDRs) of the             sampling interval T , is given by
ASPIN Laboratory were flown over on a USAF Beechcraft
C-12 Huron, a fixed-wing aircraft, to collect ambient cellular       xpva (k + 1) = Fpva xr (k)+wpva (k), k = 0, 1, 2, . . . , (1)
signals. This unique dataset consists of combinations of flight                                                  2         
runs over three different environments (rural, semi-urban, and                              I3×3 T I3×3 T2 I3×3
urban) with altitudes ranging up to 23,000 feet and a multitude                Fpva =  03×3 I3×3                T I3×3  ,
of trajectories and maneuvers including straight segments,                                 03×3 03×3              I3×3
banking turns, holding patterns, and ascending and descending                      h                  iT
                                                                                             T      T                           T
teardrops performed by members of the USAF Test Pilot               where xpva ≜ r T  r , ṙ r , r̈ r    , r r ≜ [xr , yr , zr ] is the 3-
School (TPS). During these large-scale experiments, terabytes       D position of the aircraft expressed in a North-East-Down
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                        3

(NED) frame, and wpva is a discrete-time zero-mean white              D. Altimeter Measurement Model
noise sequence with covariance Qpva given by                             Since cellular base stations appear to have similar altitudes
                    5
                         T   4
                               T 3
                                    
                                     T                                for a high-flying aircraft, their vertical dilution of precision
                         20     8     6                               (VDOP) will be very large. To circumvent this issue, altime-
                         T4    T3    T2
             Qpva =                       ⊗ S̃NED ,
                                         
                          8     3     2                               ter data zalt derived from the aircraft’s on-board navigation
                         T3    T2
                          6     2    T                                system is used in addition to the cellular carrier phase mea-
                                                                      surements in the measurement-update step in the EKF.
where ⊗ denotes the Kronecker product, and S̃NED ≜
diag [q̃N , q̃E , q̃D ], where q̃N , q̃E , and q̃D are the NED jerk
continuous-time noise power spectra, respectively. It should                         III. NAVIGATION F RAMEWORK
be noted that more complicated dynamic models can be used               This section formulates the EKF navigation framework
to describe the aircraft’s dynamics, e.g., Singer acceleration,       based on the models presented in II.
mean-adaptive acceleration, circular motion, curvilinear mo-
tion, coordinated turn, among others [30]. Of course, if an INS       A. EKF Model
is available, its measurements can be used to describe the                                                     T
aircraft’s motion, while aiding the INS with cellular SOPs [31].         Let x ≜ xT        T               T
                                                                                     pv , xclk1 , . . . , xclkN    denote the state to be
                                                                                                      h            iT
                                                                      estimated, where xclkn ≜ cδtn , cδ̇tn . Using (1) and (2),
B. Clock Error Dynamics Model                                         one can write the dynamics of x as

    Wireless standards require cellular base stations to be                             x(k + 1) = Fx(k) + w(k),                      (5)
synchronized to within a few microseconds, which is order
                                                                      where F ≜ diag [Fpva , Fclk , . . . , Fclk ] and w(k) is the overall
of magnitudes higher than the nanosecond requirements in
                                                                      process noise vector, which        is a zero-mean    white sequence
GNSS. As such, the base station clock errors, which are dy-                                                        
                                                                      with covariance Q ≜ diag Qpva , Q̄clk , and
namic and stochastic, must be accounted for in the navigation
filter when navigation with cellular SOPs. A typical model for
                                                                                                                                       
                                                                                 Qclkr+Qclks1             Qclkr      ...       Qclkr
the dynamics of the clock error states is the so-called two-state                    Qclkr          Qclkr+Qclks2 . . .        Qclkr    
model, composed of the clock bias δt and clock drift δt, ˙ given      Q̄clk ≜
                                                                              
                                                                                         ..                ..                    .      ,
                                                                                                                                        
                                                                                                                     ..          ..
by
                                                                                         .                 .            .              
            xclk (k + 1) = Fclk xclk (k) + wclk (k),          (2)                     Qclkr               Qclkr      . . . Qclkr+QclksN
                                                                                                        N
where wclk is a discrete-time zero-mean white noise sequence          where Qclkr and Qclksn n=1 have the same form as in (3),
with covariance Qclk , and                                            except that Sw̃δt and Sw̃δt˙ are replaced with the receiver and
                           "                   3         2
                                                           #          n-th base station’s clock process noise spectra, respectively.
                             Sw̃δt T +Sw̃δt˙ T3 Sw̃δt˙ T2
               
          1 T                                                         Note that the cross-correlations in Q̄clk come from combining
 Fclk =          , Qclk =                  2                 . (3)
          0 1                     Sw̃δt˙ T2      Sw̃δt˙ T             the effect of the receiver and cellular base station clocks in
                                                                      the same state. Since the receiver clock bias is common to all
The power spectra Sw̃δt and Sw̃δt˙ are determined by the quality      clock states, the cross-correlations in Q̄clk will be the receiver
of the oscillator from which the clock signal is derived [32].        clock’s process noise covariance [35].
                                                                         The measurement vector defined by z(k)                         ≜
                                                                                                        T
C. SOP Measurement Model                                              [zalt (k), z1 (k), . . . , zN (k)] is used to estimate x in the
                                                                      EKF. In vector form, the measurement equation is given by
  ASPIN Laboratory’s SDR, called MATRIX: Multichannel
Adaptive TRansceiver Information eXtractor, produces several                              z(k) = h [x(k)] + v(k),                     (6)
types of navigation observables. In order to get the highest
                                                                      where h [x(k)] is a vector-valued function defined as
possible precision, carrier phase observables are exploited for                                                                T
                                                                      h [x(k)] ≜ [halt [x(k)] , h1 [x(k)] , . . . , hN [x(k)]] with
navigation, which after some manipulations can be modeled
                                                                      halt [x(k)] = zr (k) + valt (k), hn [x(k)] ≜ ∥r r (k) − r sn ∥2 +
as [14]                                                                                                                      T
                                                                      cδtn (k), and v(k) ≜ [valt (k), v1 (k), . . . , vN (k)] is the
zn (k) = ∥r r (k) − r sn ∥2 +cδtn (k)+vn (k),  n = 1, 2, . . . , N,   measurement noise vector, which is modeled as zero-mean
                                                                (4)   white Gaussian random vector with covariance R(k) ≜
                                                                              2
where r sn is the n-th cellular base station’s 3–D position           diag σalt (k), σ12 (k), . . . , σN
                                                                                                       2
                                                                                                         (k) .
vector; c is the speed of light; δtn is the overall clock error          An EKF is implemented given the dynamics and measure-
in the n-th carrier phase measurement, which combines the             ment models in (5) and (6) to produce an estimate of x(k)
effect of receiver and base station clock biases and the initial      using all measurements up to time-step k, denoted by x̂(k|k),
carrier phase ambiguity; N is the total number of available           and an associated estimation error covariance denoted by
base stations; and vn (k) is the measurement noise, which             P(k|k). The EKF is initialized from two successive position
is modeled as a discrete-time zero-mean white Gaussian se-            priors according to the framework discussed in [35]. The EKF
quence with variance σn2 (k). The measurement noise variance          process and measurement noise covariances are described in
can be modeled as a function of the CNR [33], [34].                   the next subsection.
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                                                4

B. EKF Settings                                                        Fig. 2 shows the regions in which the experiments were
   The measurement rate was T = 0.08/3 s; the jerk process          performed as well as the aircraft trajectory for each flight. The
noise spectra were chosen to be q̃N = 18 m2 /s5 , q̃E = 18          3G base transceiver stations (BTSs) and 4G eNodeBs were
m2 /s5 , and q̃D = 0.5 m2 /s5 ; and the receiver and base station   mapped via the method described in [36]. The mapped towers
clock process noise covariance matrices were chosen to be           were cross-checked via Google Earth and online databases and
                                                                    are shown in Fig. 2. This article investigates the potential of
                      9.57 × 10−5 2.52 × 10−8
                                                   
           Qclk,r =                                   ,       (7)   cellular SOPs for navigation; therefore, mapping the SOPs will
                      2.52 × 10−8 1.89 × 10−6                       not be discussed.
                       3.11 × 10−7 2.52 × 10−11
                                                     
          Qclk,sn =                                     .     (8)
                      2.52 × 10−11 1.89 × 10−9
                                                                                                                                                     Equipment rack
The above clock process noise covariance matrices assumed             GPS antennas
the receiver to be equipped with a typical temperature-
compensate crystal oscillator (TCXO), while the cellular base
stations are equipped with a typical oven-controlled crystal
oscillator (OCXO) [8].
                                                2                                                           C-12 aircraft \Ms. Mabel"
   The altimeter measurement error variance σalt   (k) was as-
                  2                                                                                  Layer A
sumed to be 5 m . The cellular measurement noise variances                                           Layer B
                                                                                                                                 Power
were calculated as a function of the CNR and receiver param-                                         Layer C                      strip
                                                                     Cellular antennas             Power strip A                   B
eters as discussed in [33], [34]. The range of values taken by

                                                                                                                   |
                                                                      Quad-channel                                      4G @ 731.5 MHz, T-Mobile

                                                                                                                   {z
                                                                                         3 Laird antennas               or 4G @ 1955 MHz, AT&T
the measurement noise variances are explicitly stated for each        USRP-2955
                                                                                                                                                      Aircraft GPS antenna

                                                                                                                   }
                                                                                                                   |
                                                                                                                        4G @ 751MHz, Verizon
region in Section V.                                                                                                    or 4G @ 739 MHz, AT&T

                                                                                                                   {z
                                                                                                                        or 4G @ 2145 MHz, T-Mobile                 Aircraft
                                                                    Hardware setup

                                                                                                                   }
                                                                                                                                                                   navigation
                                                                                                                        3G @ 881.52 MHz, Verizon                   system
    IV. E XPERIMENTAL S ETUP AND F LIGHT R EGIONS
  This section overviews the experimental setup used for data
collection and processing. It also describes the flight regions.    Fig. 1. Hardware setup with which the C-12 aircraft was equipped.

                                                                     California, USA                                                Edwards (rural)
A. Hardware and Software Setup
   The C-12 aircraft was equipped with a universal software
radio peripheral (USRP) with consumer-grade cellular anten-
nas to sample three cellular bands and store the samples on a                                                   Region A
desktop computer for off-line processing. The stored samples                                                                        Palmdale (semi-urban)

were post-processed with the 3G and 4G cellular modules of
MATRIX. The SDR produces navigation observables: Doppler                                            Region B

frequency, carrier phase, and pseudorange, along with corre-
sponding CNRs. The hardware setup is shown in Fig. 1.
                                                                                                                                    Riverside (urban)
   The aircraft’s ground-truth trajectory was taken from the
C-12’s on-board Honeywell H764-ACE EGI INS/GPS, which                                        Region C

provided time-space-position information (TSPI) at a 1 Hz
data rate. The accuracy specifications are tabulated in Table I.

                          TABLE I
             H ONEYWELL H764-ACE EGI ACCURACY                       Fig. 2. Regions A, B, and C in which the flight campaigns took place.
                                                                    The yellow pins represent 3G and 4G cellular towers that were mapped and
         Metric         Blended INS/GPS Accuracy                    analyzed in this study. The right figures show the aircraft trajectory in all
         Position    5 m, spherical error probable (SEP)            regions (shown in red). Geographic points of interest in each region, shown
         Velocity                 0.01 m/s                          in green crosses, were chosen according to the designed trajectories.
         Heading                  0.015 deg
        Pitch/Roll                0.01 deg
                                                                                         V. A ERIAL NAVIGATION R ESULTS

B. Flight Regions                                                      This section presents experimental results demonstrating
                                                                    high-altitude aircraft navigation using the framework discussed
   Three flights are reported in this article, each of which took   in Section II in the three regions shown in Fig. 2.
place in one of three regions: (i) Region A: a rural region in
Edwards AFB, California, (ii) Region B: a semi-urban region
in Palmdale, California, and (iii) Region C: an urban region in     A. Aerial Navigation in Region A
Riverside, California. Different maneuvers were planned over          The test trajectory in Region A consisted of (i) a 24-
the three regions to test several aspects of aircraft navigation    km straight segment, followed by (ii) a 270◦ banking turn
with cellular SOPs.                                                 of length 18 km, and (iii) a final 9-km straight segment.
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                               5

The total distance traveled by the aircraft was over 51 km,
completed in 9 minutes. The aircraft maintained an altitude of
approximately 5,000 ft AGL throughout the trajectory. During
this flight, three radio frequency channels were sampled at:
(i) 881.52 MHz, which is a 3G channel allocated for the
U.S. cellular provider Verizon Wireless, (ii) 731.5 MHz, a
4G LTE channel allocated for T-Mobile, and (iii) 751 MHz,
also a 4G LTE channel allocated for Verizon. A total of 11
cellular SOPs were heard during the experiment: (a) six 3G
                                                                                                          (a)
BTSs and (b) five 4G eNodeBs. The 11 cellular SOPs were
acquired at different times and tracked for different durations
based on signal quality. Figs. 3(a)–(c) show the time history of
(i) measured CNRs, (ii) pseudorange measurements, and (iii)
pseudorange error (pseudorange minus the true range), for all
11 cellular SOPs, respectively. One can see from Fig. 3(c) that
pseudorange tracking is lost for some of the cellular SOPs at
or around 300 s, which is when the aircraft starts banking to
perform the 270◦ turn. In addition to the high dynamics of
the banking turn, it is suspected that the aircraft’s wings and
body block or severely attenuate some of the signals during
banking, causing loss of tracking. Using the expressions of the
measurement noise variances as a function of the CNR and
receiver parameters in [33], [34], σn (k) was found to vary
between 1.44 to 9.47 m.                                                                                   (b)

   Next, the state vector x of the aircraft was estimated using
the carrier phase measurements obtained from the cellular
SOP receivers via the EKF discussed in Section III-A. The
total position RMSE was calculated to be 10.5 m over the
51 km trajectory, traversed in 9 minutes. Fig. 4 shows the
aircraft’s true and estimated trajectories. Fig. 5 shows the EKF
estimation error plots and corresponding sigma-bounds for the
aircraft’s position and velocity states. It is important to note
that the position error in the EKF is the largest during the                                              (c)
turn. This is due to (i) the measurement errors due to the
                                                                   Fig. 3. (a) Time history of the CNRs for all the base stations used to compute
high dynamics of the banking turn, which severely stressed         the navigation solution in Region A. (b) Time history of the pseudoranges
the tracking loops and (ii) the mismatch in the dynamics           estimated by the cellular SOP receivers and the corresponding true range in
model assumed in the EKF, since a 270◦ banking turn has            Region A. The initial values of the pseudoranges and ranges were subtracted
                                                                   out for ease of comparison. (c) Time history of the pseudorange error
significantly different dynamics than the assumed continuous       (pseudorange minus the true range) for all cellular SOPs in Region A. The
Wiener process acceleration model. However, as mentioned           error is driven by a common term, which is the receiver’s clock bias. The
earlier, the purpose of this study is to highlight the minimum     errors increase significantly at around 300 s, which is when the turn starts.
                                                                   The high dynamics of a banking turn inject stress on the tracking loops.
performance that can be achieved with cellular SOPs. It is         The initial values of the pseudorange errors were subtracted out for ease of
important to note that the average distance between the aircraft   comparison.
and the BTSs or eNodeBs was around 30 km over the entire
trajectory, with eNodeB 4 being tracked at a 100 km distance
in the first part of the trajectory.                               experiment: (a) nine 3G BTSs and (b) five 4G eNodeBs. The
                                                                   14 cellular SOPs were acquired at different times and tracked
                                                                   for different durations based on signal quality. Figs. 6(a)–(c)
B. Aerial Navigation in Region B                                   show the time history of (i) measured CNRs, (ii) pseudorange
   The test trajectory in Region B consisted of (i) an approach    measurements, and (iii) pseudorange error (pseudorange minus
to William J Fox Airport, followed by (ii) a touch and             the true range), for all 14 cellular SOPs, respectively. Using the
go. The total distance traveled by the aircraft was over 57        expressions of the measurement noise variances as a function
km completed in 11 minutes. The aircraft descended from            of the CNR and receiver parameters in [33], [34], σn (k) was
an altitude of 7,000 ft AGL. During this flight, three radio       found to vary between 1.3 to 4.43 m.
frequency channels were sampled at: (i) 881.52 MHz, which             Next, the state vector x of the aircraft was estimated using
is a 3G channel allocated for the U.S. cellular provider Verizon   the carrier phase measurements obtained from the cellular
Wireless, (ii) 731.5 MHz, a 4G LTE channel allocated for T-        SOP receivers via the EKF discussed in Section III-A. The
Mobile, and (iii) 739 MHz, also a 4G LTE channel allocated         total position RMSE was calculated to be 4.95 m over the
for AT&T. A total of 14 cellular SOPs were heard during the        57 km trajectory, traversed in 11 minutes. Fig. 7 shows the
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                           6

     Region A at 5,000 ft AGL

            3G @ 881.52 MHz
            4G @ 731.50 MHz
            4G @ 751.00 MHz

                                                                      km
                                                                                     Fig. 5. EKF plots showing the time history of the position and velocity

                                                                   20
                                                                                     errors in Region A as well as the ±3σ bounds. As expected, the EKF
                                                                                     performs poorly in the second leg, where the mismatch between the true
                                                                                     aircraft dynamics and the assumed EKF model is highest.

                                                                                     sixth minute as the aircraft is flying on one side of the SOPs.
                                                                                     This explains the increasing sigma-bounds in Fig. 8.

                                                3
                                                                                     C. Aerial Navigation in Region C
                                                                                        The test trajectory in Region C consisted of a holding
                                                                                     pattern over Riverside Municipal Airport. The total distance
                                                                                     traveled by the aircraft was over 55 km, completed in 8.5
            2                                                                        minutes. The aircraft maintained an altitude of approximately
                                                  1.1 km                             15,000 ft AGL throughout the trajectory. During this flight,
                                                                                     two radio frequency channels were sampled at: (i) 881.52
                                                                                     MHz, which is a 3G channel allocated for the U.S. cellular
                                                                                     provider Verizon Wireless, (ii) 1955 MHz, a 4G LTE channel
                                                                                     allocated for AT&T, and (iii) (ii) 2145 MHz, a 4G LTE
                                                                                     channel allocated for T-Mobile. A total of 11 cellular SOPs
                                         10.4 m                                      were heard during the experiment: (a) seven 3G BTSs and
                                                                                     (b) four 4G eNodeBs. The 11 cellular SOPs were acquired
                                                                                     at different times and tracked for different durations based
                                                                                     on signal quality. Figs. 9(a)–(c) show the time history of
                                1                                                    (i) measured CNRs, (ii) pseudorange measurements, and (iii)
                                                            Ground-truth
                                                                                     pseudorange error (pseudorange minus the true range), for all
                                                            Estimated                9 cellular SOPs, respectively. Similar to the first flight, one
                                                                                     can see from Fig. 9(c) that pseudorange tracking is lost for
                                                                                     some of the cellular SOPs when the aircraft starts banking to
                                                                                     perform the turns in the holding pattern. Using the expressions
                                                                                     of the measurement noise variances as a function of the CNR
Fig. 4. Experimental layout and results in Region A showing: (i) BTS                 and receiver parameters in [33], [34], σn (k) was found to vary
and eNodeB positions, (ii) true aircraft trajectory, and (iii) aircraft trajectory
estimated exclusively using cellular SOPs. The aircraft traversed a total            between 1.73 to 5.69 m.
distance of 51 km traversed in 9 minutes during the experiment. The position            Next, the state vector x of the aircraft was estimated using
RMSE over the entire trajectory was found to be 10.5 m.                              the carrier phase measurements obtained from the cellular SOP
                                                                                     receivers via the EKF discussed in Section III-A. The total
                                                                                     position RMSE was calculated to be 15.44 m over the 55
aircraft’s true and estimated trajectories. Fig. 8 shows the EKF                     km trajectory, traversed in 8.5 minutes. Fig. 10 shows the
estimation error plots and corresponding sigma-bounds for the                        aircraft’s true and estimated trajectories. Fig. 11 shows the
aircraft’s position and velocity states. It is important to note                     EKF estimation error plots and corresponding sigma-bounds
that the aircraft’s position estimate on touchdown is less than 3                    for the aircraft’s position and velocity states. As expected, the
m away from the true position and is well within the runway.                         measurement errors and the mismatch in the dynamics model
In addition, the geometric diversity becomes poor after the                          assumed in the EKF are more severe during the turns.
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                                 7

                                                                                                                             Region B at 0 ft { 7,000 AGL

                                                                                                                                         3G @ 881.52 MHz
                                                                                                                                         4G @ 731.50 MHz
                                                                                                                                         4G @ 739.00 MHz

                                       (a)

                                                                                                            20 km

                                                                                                                                             Ground-truth
                                                                                               3                                             Estimated

                                                                                                           2               4
                                       (b)

                                                                                                           7                        6

                                                                                                                            5
                                                                                                 Touch down

                                                                                                   2.9 m
                                       (c)

Fig. 6. (a) Time history of the CNRs for all the base stations used to compute
the navigation solution in Region B. (b) Time history of the pseudoranges
estimated by the cellular SOP receivers and the corresponding true range in                                                                      1
Region B. The initial values of the pseudoranges and ranges were subtracted
out for ease of comparison. (c) Time history of the pseudorange error
(pseudorange minus the true range) for all cellular SOPs in Region B.            Fig. 7. Experimental layout and results in Region B showing: (i) BTS
                                                                                 and eNodeB positions, (ii) true aircraft trajectory, and (iii) aircraft trajectory
                                                                                 estimated exclusively using cellular SOPs. The aircraft traversed a total
                                                                                 distance of 57 km traversed in 11 minutes during the experiment. The position
D. Discussion                                                                    RMSE over the entire trajectory was found to be 4.96 m. Note that the position
                                                                                 estimate on touchdown is less than 3 m away from the true aircraft position
                                                                                 and is well within the runway.
   The navigation performance in all three Regions is summa-
rized in Table II.
                                                                                    The achieved results unveiled the remarkable potential of
                                 TABLE II                                        utilizing cellular SOPs for sustained accurate high altitude
         NAVIGATION P ERFORMANCE WITH C ELLULAR S IGNALS
                                                                                 aircraft navigation. The results presented herein, although
           Metric           Region A              Region B       Region C        promising, can be further improved upon in several ways.
 Cellular towers {3G, 4G}    {6, 5}                 {9, 5}        {7, 4}         The following are key takeaways and design consideration for
 Cellular frequencies (MHz)  881.52                881.52         881.52         reliable aircraft navigation with cellular SOPs.
                             731.5                  731.5          1955
                              751                    739           2145             • Accounting for the aircraft dynamical model mismatch:
   Flight duration (mins)       9                     11            8.5                Aircraft, such as the C-12, can perform a variety of highly
      Flight length (km)       51                     57             55                dynamic maneuvers. The dynamics model employed in
      Altitude AGL (ft)      5,000                0 – 7,000       15,000               the EKF in this study did not perfectly capture the
     Position RMSE (m)       10.53                   4.96          11.67               aircraft dynamics throughout its trajectory, leading to
   Velocity RMSE (m/s)        0.58                   0.50           0.71               increased estimation error due to the mismatch between
Maximum position error (m) 22.67                    15.04         25.89                the actual aircraft’s dynamics and the dynamical model
Maximum velocity error (m/s) 2.29                    3.19           3.94               assumed by the EKF. This mismatch can be mitigated
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                             8

                                                                                                                        (a)

Fig. 8. EKF plots showing the time history of the position and velocity errors
in Region B as well as the ±3σ bounds.

       by using appropriate dynamical models for fixed-wing
       aircraft or more elaborate dynamical models (e.g., Wiener
       process acceleration, Singer acceleration, mean-adaptive
       acceleration, semi-Markov jump process, circular motion,
       curvilinear motion, coordinated turn, among others [30])                                                         (b)
       coupled with adaptive estimation techniques [37]–[42].
       Alternatively, if access to raw IMU data is available, a
       kinematic model with IMU measurements can be used,
       as is the case with most INS aiding techniques [10], [31].
   •   Accounting for statistical model mismatch: The aircraft’s
       process noise covariance assumed by the EKF’s dy-
       namical model was found via off-line tuning and by
       analyzing the aircraft’s maneuvers from ground truth data.
       In addition, the process noise covariances of the aircraft’s
       receiver clock was set at typical TCXO values and the                                                            (c)
       cellular SOP transmitter clocks were set at typical OCXO                  Fig. 9. (a) Time history of the CNRs for all the base stations used to compute
       values. While these values represent good approximations                  the navigation solution in Region C. (b) Time history of the pseudoranges
       for the aircraft’s receiver clock quality as well as the                  estimated by the cellular SOP receivers and the corresponding true range in
                                                                                 Region C. The initial values of the pseudoranges and ranges were subtracted
       quality of typical cellular SOP transmitters, mismatches                  out for ease of comparison. (c) Time history of the pseudorange error
       between the assumed values and the actual values can                      (pseudorange minus the true range) for all cellular SOPs in Region C. The
       be mitigated via adaptive estimation techniques [43]–                     error is driven by a common term, which is the receiver’s clock bias.
       [45], which would improve the estimation performance.
       Adaptive estimation techniques would also mitigate the
       errors arising from mismatches between the actual mea-                          the radio simultaneous localization and mapping (radio
       surement noise variances and calculated measurement                             SLAM) framework, which maps the unknown SOPs
       noise variances.                                                                simultaneously with localizing the aircraft [14], [31].
   •   Vertical dilution of precision: At high altitudes, there
       is very little vertical diversity with respect to terrestrial                                        VI. C ONCLUSION
       cellular towers. As such, the aircraft’s cellular-based                      This article demonstrated robust high altitude aircraft nav-
       navigation solution VDOP will be large. Nevertheless,                     igation with 3G CDMA and 4G LTE cellular SOPs. An
       the aircraft’s vertical position can still be estimated from              EKF was used to fuse cellular carrier phase measurements to
       the pseudoranges extracted from cellular towers, albeit                   estimate the aircraft’s position, velocity, and time. The EKF
       with less accuracy compared to the results presented in                   utilized a simple, yet effective continuous Wiener process ac-
       this paper, which fused altimeter-based measurements.                     celeration model to describe the aircraft dynamics. A multitude
   •   Mapping cellular SOPs: This article assumed cellular                      of flight trajectories and altitudes above ground level (AGL)
       SOPs to be mapped a priori. This was achieved via a                       were exercised in the three flights: (i) a 51 km trajectory of
       mapping campaign according to the method described in                     grid maneuvers with banking and straight segments at about
       [36]. Nevertheless, such assumption can be relaxed via                    5,000 ft AGL, (ii) a 57 km trajectory of a teardrop descent
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                                                                                               9

                                                                                                                                                 Error                                 3 bounds
  Region C at 15,000 ft AGL

                                                                                      North position error (m)
                                                                                                                   20                                                                  20

                                                                                                                                                           East position error (m)
                                                                                                                   10                                                                  10

                                                                                                                    0                                                                   0

                                                                                                                   -10                                                                 -10

                                                                                                                   -20                                                                 -20
                                                                                                                         0   100   200   300   400   500                                     0   100   200   300   400   500
                                                                                                                                    Time (s)                                                            Time (s)

                                                                                      North velocity error (m/s)

                                                                                                                                                           East velocity error (m/s)
                                                                                                                   20                                                                  20
                       3G @ 881.52 MHz
                       4G @ 1955 MHz                                                                               10                                                                  10
                       4G @ 2145 MHz
                                                                                                                    0                                                                   0

                                                 km                                                                -10                                                                 -10

                                            20                                                                     -20                                                                 -20
                                                                                                                         0   100   200   300   400   500                                     0   100   200   300   400   500
                                                                                                                                    Time (s)                                                            Time (s)

                                                                                     Fig. 11. EKF plots showing the time history of the position and velocity
                                                                                     errors in Region C as well as the ±3σ bounds. As expected, the EKF
                                                                                     performs poorly in the second leg, where the mismatch between the true
                                                                                     aircraft dynamics and the assumed EKF model is highest.

                                                                                                        ACKNOWLEDGMENTS
                                                 2                                      The authors would like to thank Edwards AFB and Hol-
                               6                                                     loman AFB for inviting the ASPIN Laboratory to conduct
                                                                                     experiments on Air Force aircraft in the “SNIFFER: Sig-
                                                                                     nals of opportunity for Navigation In Frequency-Forbidden
                                                                                     EnviRonments” flight campaign. The authors would like to
          5                                                             3            thank Joshua Morales, Kimia Shamaei, Mahdi Maaref, Kyle
                                                                                     Semelka, MyLinh Nguyen, and Trier Mortlock for their
                                                                                     help with preparing for data collection. DISTRIBUTION
                                                                                     STATEMENT A. Approved for public release; Distribution
                                             1                                       is unlimited. 412TW-PA-20146.

                                                        7                                                                                       R EFERENCES
                                                                                       [1] ASRS/NASA, “ASRS Database Online,” https://asrs.arc.nasa.gov/
                                                                                           search/database.html.
                eNodeB 3
                                                                                       [2] M. Harris, “FAA files reveal a surprising threat to airline safety,” IEEE
                                                              Ground-truth                 Spectrum Online. https://spectrum.ieee.org/faa-files-reveal-a-surprising-
                                                                                           threat-to-airline-safety-the-us-militarys-gps-tests, January 2021.
               4                                              Estimated
                                                                                       [3] D. Miralles, A. Bornot, P. Rouquette, N. Levigne, D. Akos, Y.-H.Chen,
                                                                                           S. Lo, and T. Walter, “An assessment of GPS spoofing detection via
                                                                                           radio power and signal quality monitoring for aviation safety operations,”
                                                                                           IEEE Intelligent Transportation Systems Magazine, vol. 12, no. 3, pp.
                                                                                           136–146, June 2020.
                                                                                       [4] F. Dovis, L. Ruotsalainen, R. Toledo-Moreo, Z. Kassas, and V. Gikas,
Fig. 10. Experimental layout and results in Region C showing: (i) BTS                      “Recent advancement on the use of global navigation satellite system-
and eNodeB positions, (ii) true aircraft trajectory, and (iii) aircraft trajectory         based positioning for intelligent transport systems,” IEEE Intelligent
estimated exclusively using cellular SOPs. The aircraft traversed a total                  Transportation Systems Magazine, vol. 12, no. 3, pp. 6–9, June 2020.
distance of 55 km traversed in 8.5 minutes during the experiment. The position         [5] International Civil Aviation Organization (ICAO), “An urgent
RMSE over the entire trajectory was found to be 15.44 m.                                   need to address harmful interferences to GNSS,” https:
                                                                                           //www.iata.org/contentassets/e45e5219cc8c4277a0e80562590793da/
                                                                                           address-harmful-interferences-gnss.pdf, Tech. Rep., May 2019.
                                                                                       [6] A. Hansen, S. Mackey, H. Wassaf, E. Wallischeck, C. Scarpone,
from 7,000 ft AGL down to touchdown at the runway, and                                     M. Barzach, and E. Baskerville, “Complementary PNT and GPS backup
(iii) a 55 km trajectory of a holding pattern at about 15,000 ft                           technologies demonstration report: Sections 1 through 10,” John A. Volpe
                                                                                           National Transportation Systems Center (US), no. DOT-VNTSC-20-07,
AGL. Cellular SOPs produced remarkable navigation accuracy                                 2021.
in all three flights, achieving a 3-D position RMSE of 10.53                           [7] M. Bartock, J. Brule, Y. Li-Baboud, S. Lightman, J. McCarthy,
m, 4.96 m, and 15.44 m, respectively. These unprecedented                                  K. Reczek, D. Northrip, A. Scholz, and T. Suloway, “Foundational PNT
                                                                                           profile: Applying the cybersecurity framework for the responsible use of
results demonstrate the potential of cellular signals as a viable                          positioning, navigation, and timing (PNT) services,” National Institute
alternative to GNSS for high altitude aircraft navigation.                                 of Standards and Technology (NIST), no. NISTIR 8323, February 2021.
While the presented are encouraging, more accurate navigation                          [8] Z. Kassas, “Position, navigation, and timing technologies in the 21st
                                                                                           century,” J. Morton, F. van Diggelen, J. Spilker, Jr., and B. Parkinson,
results can be achieved by fusing cellular SOP observables                                 Eds. Wiley-IEEE, 2021, vol. 2, ch. 38: Navigation with Cellular Signals
with an INS.                                                                               of Opportunity, pp. 1171–1223.
Flight Demonstration of High Altitude Aircraft Navigation with Cellular Signals
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                            10

 [9] A. Abdallah, J. Khalife, and Z. Kassas, “Experimental characterization     [29] Z. Kassas, J. Khalife, A. Abdallah, C. Lee, J. Jurado, S. Wachtel,
     of received 5G signals carrier-to-noise ratio in indoor and urban envi-         J. Duede, Z. Hoeffner, T. Hulsey, R. Quirarte, and R. Tay, “Assessment
     ronments,” in Proceedings of IEEE Vehicular Technology Conference,              of cellular signals of opportunity for high altitude aircraft navigation,”
     April 2021, pp. 1–5.                                                            IEEE Aerospace and Electronic Systems Magazine, vol. 37, no. 10, pp.
[10] Z. Kassas, M. Maaref, J. Morales, J. Khalife, and K. Shamaei, “Robust           4–19, October 2022, doi: 10.1109/MAES.2022.3187142.
     vehicular localization and map matching in urban environments through      [30] X. Li and V. Jilkov, “Survey of maneuvering target tracking.
     IMU, GNSS, and cellular signals,” IEEE Intelligent Transportation               Part I: Dynamic models,” IEEE Transactions on Aerospace and
     Systems Magazine, vol. 12, no. 3, pp. 36–52, June 2020.                         Electronic Systems, vol. 39, no. 4, pp. 1333–1364, 2003, doi:
[11] M. Maaref and Z. Kassas, “Autonomous integrity monitoring for vehic-            10.1109/TAES.2003.1261132.
     ular navigation with cellular signals of opportunity and an IMU,” IEEE     [31] J. Morales and Z. Kassas, “Tightly-coupled inertial navigation system
     Transactions on Intelligent Transportation Systems, vol. 23, no. 6, pp.         with signals of opportunity aiding,” IEEE Transactions on Aerospace
     5586–5601, June 2022.                                                           and Electronic Systems, vol. 57, no. 3, pp. 1930–1948, 2021.
[12] J. Khalife and Z. Kassas, “On the achievability of submeter-               [32] J. Barnes, A. Chi, R. Andrew, L. Cutler, D. Healey, D. Leeson,
     accurate UAV navigation with cellular signals exploiting loose net-             T. McGunigal, J. Mullen, W. Smith, R. Sydnor, R. Vessot, and G. Win-
     work synchronization,” IEEE Transactions on Aerospace and Elec-                 kler, “Characterization of frequency stability,” IEEE Transactions on
     tronic Systems, vol. 58, no. 5, pp. 4261–4278, October 2022, doi:               Instrumentation and Measurement, vol. 20, no. 2, pp. 105–120, May
     10.1109/TAES.2022.3162770.                                                      1971.
[13] J. Khalife and Z. Kassas, “Differential framework for submeter-accurate    [33] J. Khalife, K. Shamaei, and Z. Kassas, “Navigation with cellular CDMA
     vehicular navigation with cellular signals,” IEEE Transactions on Intel-        signals – part I: Signal modeling and software-defined receiver design,”
     ligent Vehicles, 2022, accepted, doi: 10.1109/TIV.2022.3187957.                 IEEE Transactions on Signal Processing, vol. 66, no. 8, pp. 2191–2203,
[14] Z. Kassas, J. Khalife, A. Abdallah, and C. Lee, “I am not afraid                April 2018.
     of the GPS jammer: resilient navigation via signals of opportu-            [34] K. Shamaei and Z. Kassas, “LTE receiver design and multipath analysis
     nity in GPS-denied environments,” IEEE Aerospace and Electronic                 for navigation in urban environments,” NAVIGATION, Journal of the
     Systems Magazine, vol. 37, no. 7, pp. 4–19, July 2022, doi:                     Institute of Navigation, vol. 65, no. 4, pp. 655–675, December 2018.
     10.1109/MAES.2022.3154110.                                                 [35] J. Khalife and Z. Kassas, “Opportunistic UAV navigation with carrier
[15] Y. Zeng, Q. Wu, and R. Zhang, “Accessing from the sky: A tutorial on            phase measurements from asynchronous cellular signals,” IEEE Trans-
     UAV communications for 5G and beyond,” Proceedings of the IEEE,                 actions on Aerospace and Electronic Systems, vol. 56, no. 4, pp. 3285–
     vol. 107, no. 12, pp. 2327–2375, December 2019.                                 3301, August 2020, doi: 10.1109/TAES.2019.2948452.
[16] X. Cai, J. Rodriguez-Pineiro, X. Yin, N. Wang, B. Ai, G. Pedersen, and     [36] J. Morales and Z. Kassas, “Optimal collaborative mapping of terres-
     A. Yuste, “An empirical air-to-ground channel model based on passive            trial transmitters: receiver placement and performance characterization,”
     measurements in LTE,” IEEE Transactions on Vehicular Technology,                IEEE Transactions on Aerospace and Electronic Systems, vol. 54, no. 2,
     vol. 68, no. 2, pp. 1140–1154, February 2019.                                   pp. 992–1007, April 2018.
[17] W. Khawaja, I. Guvenc, D. Matolak, U. Fiebig, and N. Schneckenburger,      [37] M. Yeddanapudi, Y. Bar-Shalom, and K. Pattipati, “IMM estimation
     “A survey of air-to-ground propagation channel modeling for unmanned            for multitarget-multisensor air traffic surveillance,” Proceedings of the
     aerial vehicles,” IEEE Communications Surveys & Tutorials, vol. 21,             IEEE, vol. 85, no. 1, pp. 80–96, January 1997.
     no. 3, pp. 2361–2391, 2019, doi: 10.1109/COMST.2019.2915069.               [38] A. Mohamed and K. Schwarz, “Adaptive Kalman filtering for
[18] K. Matheou, R. Apaza, A. Downey, R. Kerczewski, J. Jung, C. Ippolito,           INS/GPS,” Journal of Geodesy, vol. 73, no. 4, pp. 193–203, 1999.
     and H. Modi, “Analysis of at-altitude LTE power spectra for small          [39] P. Hanlon and P. Maybeck, “Multiple-model adaptive estimation us-
     unmanned aircraft system C2 communications,” in Proceedings of                  ing a residual correlation Kalman filter bank,” IEEE Transactions on
     Integrated Communications, Navigation and Surveillance Conference,              Aerospace and Electronic Systems, vol. 36, no. 2, pp. 393–406, April
     April 2019, pp. 1–12.                                                           2000.
[19] X. Cai, N. Wang, J. Rodriguez-Pineiro, X. Yin, A. Yuste, W. Fan,           [40] C. Hide, T. Moore, and M. Smith, “Adaptive Kalman filtering for low-
     G. Zhang, G. Pedersen, and L. Tian, “Low altitude air-to-ground                 cost INS/GPS,” The Journal of Navigation, vol. 56, no. 1, pp. 143–152,
     channel characterization in LTE network,” in Proceedings of European            January 2003.
     Conference on Antennas and Propagation, April 2019, pp. 1–5.               [41] W. Ding, J. Wang, C. Rizos, and D. Kinlyside, “Improving adaptive
[20] B. Van Der Bergh, A. Chiumento, and S. Pollin, “LTE in the sky: trading         Kalman estimation in GPS/INS integration,” The Journal of Navigation,
     off propagation benefits with interference costs for aerial nodes,” IEEE        vol. 60, no. 3, pp. 517–529, September 2007.
     Communications Magazine, vol. 54, no. 5, pp. 44–50, May 2016.              [42] M. Yu, “INS/GPS integration system using adaptive filter for estimating
[21] I. Kovacs, R. Amorim, H. Nguyen, J. Wigard, and P. Mogensen,                    measurement noise variance,” IEEE Transactions on Aerospace and
     “Interference analysis for UAV connectivity over LTE using aerial               Electronic Systems, vol. 48, no. 2, pp. 1786–1792, April 2012.
     radio measurements,” in Proceedings of IEEE Vehicular Technology           [43] Z. Kassas, V. Ghadiok, and T. Humphreys, “Adaptive estimation of
     Conference, September 2017, pp. 1–6.                                            signals of opportunity,” in Proceedings of ION GNSS Conference,
[22] R. Amorim, J. Wigard, I. Kovacs, and T. Sorensen, “UAV communica-               September 2014, pp. 1679–1689.
     tions for 5G and beyond,” Y. Zeng, I. Guvenc, R. Zhang, G. Geraci, and     [44] J. Dunik, , O. Straka, O. Kost, and J. Havlik, “Noise covariance
     D. Matolak, Eds. Wiley-IEEE, 2021, ch. 5: Performance Enhancements              matrices in state-space models: A survey and comparison of estimation
     for LTE-Connected UAVs: Experiments and Simulations, pp. 139–161.               methods—part i,” International Journal of Adaptive Control and Signal
[23] E. Teng, J. Diogo Falcao, and B. Iannucci, “Holes-in-the-sky: A field           Processing, vol. 31, no. 11, pp. 1505–1543, May 2017.
     study on cellular-connected UAS,” in Proceedings of International          [45] J. Dunik, O. Kost, O. Straka, and E. Blasch, “Navigation and estimation
     Conference on Unmanned Aircraft Systems, June 2017, pp. 1165–1174.              improvement by environmental-driven noise mode detection,” in Pro-
[24] X. Lin, V. Yajnanarayana, S. Muruganathan, S. Gao, H. Asplund,                  ceedings of IEEE/ION Position, Location, and Navigation Symposium,
     H. Maattanen, M. Bergstrom, S. Euler, and Y. Wang, “The sky is not              2020, pp. 925–932.
     the limit: LTE for unmanned aerial vehicles,” IEEE Communications
     Magazine, vol. 56, no. 4, pp. 204–210, April 2018.
[25] G. Athanasiadou, M. Batistatos, D. Zarbouti, and G. Tsoulos, “LTE
     ground-to-air field measurements in the context of flying relays,” IEEE
     Wireless Communications, vol. 26, no. 1, pp. 12–17, February 2019.
[26] A. Abdalla and V. Marojevic, “Communications standards for unmanned
     aircraft systems: The 3GPP perspective and research drivers,” IEEE
     Communications Standards Magazine, vol. 5, no. 1, pp. 70–77, March
     2021.
[27] H. Maattanen, “UAV communications for 5G and beyond,” Y. Zeng,
     I. Guvenc, R. Zhang, G. Geraci, and D. Matolak, Eds. Wiley-IEEE,
     2021, ch. 6: 3GPP Standardization for Cellular-Supported UAVs, pp.
     163–180.
[28] E. Kim and Y. Shin, “Feasibility analysis of LTE-based UAS navigation
     in deep urban areas and DSRC augmentation,” Sensors, vol. 19, no. 9,
     pp. 4192–4207, April 2019, doi: 10.3390/s19194192.
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                                                                                                             11

                          Zaher (Zak) M. Kassas (zkassas@ieee.org) is a                                   Juan Jurado is a U.S. Air Force Lieutenant Colonel
                          professor at The Ohio State University and direc-                               and the Director of Education at the U.S. Air Force
                          tor of the Autonomous Systems Perception, Intelli-                              Test Pilot School. He holds a B.S. from Texas A&M
                          gence, and Navigation (ASPIN) Laboratory. He is                                 University, a M.S. from the Air Force Test Pilot
                          also director of the U.S. Department of Transporta-                             School, and M.S. and Ph.D. from the Air Force
                          tion Center: CARMEN (Center for Automated Vehi-                                 Institute of Technology. Previously, he served as
                          cle Research with Multimodal AssurEd Navigation),                               Director of Engineering for the 413th Flight Test
                          focusing on navigation resiliency and security of                               Squadron and oversaw various C-130, V-22, and H-
                          highly automated transportation systems. He re-                                 1 flight test programs. His research interests include
                          ceived a B.E. in Electrical Engineering from the                                aircraft performance modeling, online sensor cali-
                          Lebanese American University, an M.S. in Electrical                             bration, image processing, visual-inertial navigation,
and Computer Engineering from The Ohio State University, and an M.S.E. in        and statistical sensor management for multi-sensor navigation problems.
Aerospace Engineering and a Ph.D. in Electrical and Computer Engineering
from The University of Texas at Austin. He is a recipient of the 2018 Na-
tional Science Foundation (NSF) Faculty Early Career Development Program
(CAREER) award, 2019 Office of Naval Research (ONR) Young Investigator
Program (YIP) award, 2022 Air Force Office of Scientific Research (AFOSR)
YIP award, 2018 IEEE Walter Fried Award, 2018 Institute of Navigation                                     Steven Wachtel is a U.S. Air Force Captian and
(ION) Samuel Burka Award, and 2019 ION Col. Thomas Thurlow Award.                                         a Flight Test Engineer, assigned to the 780th Test
He is a Senior Editor of the IEEE Transactions on Intelligent Vehicles and                                Squadron, Eglin AFB, FL. He received a B.S. in
an Associate Editor of the IEEE Transactions on Aerospace and Electronic                                  Mechanical Engineering from The Ohio State Uni-
Systems and the IEEE Transactions on Intelligent Transportation Systems.                                  versity, an M.S. in Flight Test Engineering from the
His research interests include cyber-physical systems, estimation theory, nav-                            U.S. Air Force Test Pilot School, and an M.S. in
igation systems, autonomous vehicles, and intelligent transportation systems.                             Systems Engineering from the Air Force Institute of
                                                                                                          Technology.

                          Joe Khalife (S’15-M’20) is a postdoctoral fellow at
                          the University of California, Irvine and member of
                          the Autonomous Systems Perception, Intelligence,
                          and Navigation (ASPIN) Laboratory. He received a                               Jacob Duede is a Major in the U.S. Air Force. He
                          B.E. in Electrical Engineering, an M.S. in Computer                            was trained as a Communication/Navigation/Mission
                          Engineering from the Lebanese American University                              Systems apprentice on C-17 Globmaster II aircraft
                          and a Ph.D. in Electrical Engineering and Computer                             and stationed at McChord Air Force Base, WA. He
                          Science from the University of California, Irvine.                             graduated from the U.S. Air Force Academy as a
                          From 2012 to 2015, he was a research assistant                                 commissioned officer with a B.S. in Mechanical
                          at LAU, and has been a member of the ASPIN                                     Engineering. He attended the Undergraduate Pilot
                          Laboratory since 2015. He is a recipient of the 2016                           Training at Columbus Air Force Base, MS. In 2020,
IEEE/ION Position, Location, and Navigation Symposium (PLANS) Best                                       he graduated from the U.S. Air Force Test Pilot
Student Paper Award and the 2018 IEEE Walter Fried Award. His research                                   School at Edwards Air Force Base, CA. He is a
interests include opportunistic navigation, autonomous vehicles, and software-                           Senior Pilot with over 2,000 hours and holds an M.S.
defined radio.                                                                   in Engineering from the University of Arkansas and an M.S. in Flight Test
                                                                                 Engineering from Air University.

                         Ali A. Abdallah (S’13) is a Ph.D. student in the
                         Department of Electrical Engineering and Computer
                                                                                                          Zachary Hoeffner is a flight test engineer at the
                         Science at the University of California, Irvine and
                                                                                                          U.S. Air Force. He received a a B.S in Nuclear En-
                         a member of the Autonomous Systems Perception,
                                                                                                          gineering from the U.S. Air Force Academy, an M.S.
                         Intelligence, and Navigation (ASPIN) Laboratory.
                                                                                                          in Flight Test Engineering from the U.S. Air Force
                         He received a B.E. in Electrical Engineering from
                                                                                                          Test Pilot School, an M.S. in Engineering Physics
                         the Lebanese American University. He is a recip-
                                                                                                          and Applied Physics from the Air Force Institute
                         ient of the Best Student Paper Award at the 2020
                                                                                                          of Technology, and an M.S. in Nuclear Engineering
                         IEEE/ION Position, Location, and Navigation Sym-
                                                                                                          from the Air Force Institute of Technology.
                         posium (PLANS).

                         Chiawei Lee is an Assistant Professor and Instructor
                         Flight Test Engineer at the U.S. Air Force Test                                  Thomas Hulsey is a U.S. Air Force Flight Comman-
                         Pilot School. He serves as the Test Management                                   der of Operations Engineering. He received a B.S.
                         Program Director where he oversees about a dozen                                 in Aerospace Engineering from Missouri University
                         student and staff led flight test projects each year.                            of Science and Technology, an M.S. in Aeronautical
                         In addition, he is the Chief Test Safety Officer                                 Engineering from the Air Force Institute of Tech-
                         responsible for the safe execution of curriculum and                             nology, and an M.S. in Experimental Flight Test
                         flight test project safety packages. He received a                               Engineering from the United States Air Force Test
                         B.S. in Aerospace Engineering from University of                                 Pilot School.
                         California, Los Angeles and a M.S. in Aero/Astro
                         Engineering from Stanford University.
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE                            12

                   Rachel Quirarte is a KC-46 and KC-135 program-
                   matic flight commander and test pilot in the 418th
                   Flight Test Squadron in the U.S. Air Force. She
                   received a a B.S. in Aeronautical Engineering from
                   the U.S. Air Force Academy, an M.S. in Flight
                   Test Engineering from the U.S. Air Force Test Pilot
                   School, and an M.S. in Mechanical Engineering
                   from Rice University.

                   RunXuan Tay received a B.S. degree in Electrical
                   Engineering from the University California, San
                   Diego and M.S. degree in Flight Test Engineering
                   from the U.S. Air Force Test Pilot School. He is
                   currently a test pilot at Air Warfare Center, Republic
                   of Singapore Air Force, where he works on fixed
                   wing test programs.
You can also read